🐛 Parse resp-text with invalid resp-text-code
#601
Merged
+40
−2
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This can be used to catche parse errors, restore the parser state, and then either call a fallback or re-raise the error. NOTE: Reckless backtracking can lead to O(n**2) situations, so it should very rarely be used. Fallbacks should not backtrack.
`resp-text` is defined as: ```abnf resp-text = ["[" resp-text-code "]" SP] text ; RFC3501 resp-text = ["[" resp-text-code "]" SP] [text] ; RFC9051 ``` And our even more lenient interpretation (which incorporates RFC9051 errata), is this: ```abnf resp-text = ["[" resp-text-code "]" [SP [text]] / [text] ``` And, although the `resp-text-code` grammar is elaborate, the final alternative is a superset of every other alternative (in both RFCs): ```abnf resp-text-code /= atom [SP 1*<any TEXT-CHAR except "]">] ``` Notice that `text` is a superset of `atom`, `SP`, `"["`, `"]"`, and every other allowed character in `resp-text-code`. So, even if `resp-text-code` fails, `resp-text` may still succeed by parsing everything as `text`. However, (prior to this commit) the parser commits to `resp-text-code` as soon as `"["` is encountered at the beginning of `resp-text`. If what follows is not `resp-text-code`, maybe it doesn't begin with an `atom` or it doesn't have a closing `"]"`, then we fail to parse `resp-text` correctly. Fixing this requires either looking further ahead more than a single token or backtracking when `resp-text-code` fails to parse. In this case, I think backtracking is the best approach. We don't need to worry about backtracking taking exponential time, because the fallback (`text`) is exceptionally simple (does not call any other productions), will parse all the way up to the next CRLF, and so nothing backtrackable can be nested. This uses `ResponseParser#current_state` that was added for #599 and relies on #600 to ensure that debug error messages are not printed when backtracking. Fixes #597.
nevans
force-pushed
the
parser/backtracking-in-resp-text
branch
from
675b663 to
7fb8b57
Compare
Collaborator
Author
|
Note that this still preserves
|
Collaborator
Author
|
Another follow-up PR is still needed to handle On the other hand, this might be considered a breaking change: it would break code that expects |
nevans
commented
Feb 11, 2026
| rescue ResponseParseError => error | ||
| raise if /\buid-set\b/i.match? error.message | ||
| restore_state state | ||
| text |
Collaborator
Author
There was a problem hiding this comment.
Oops. This is a bug. It should be ResponseText.new(nil, text? || "")
nevans
added a commit
that referenced
this pull request
Feb 11, 2026
This bug was introduced in #601, which hasn't been released yet.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
resp-textis defined as:And our even more lenient interpretation (which incorporates RFC9051 errata), is this:
And, although the
resp-text-codegrammar is elaborate, in both RFC3501 and RFC9051, the final alternative is a superset of every other alternative:Notice that
textis a superset ofatom,SP,"[","]", and every other allowed character inresp-text-code. So, ifresp-text-codefails,resp-textshould backtrack and fallback to just parsing everything astext.However, (prior to this PR) the parser commits to
resp-text-codeas soon as"["is encountered at the beginning ofresp-text. If what follows is notresp-text-code, maybe it doesn't begin with anatomor it doesn't have a closing"]", then we fail to parseresp-textcorrectly.Fixing this requires either looking further ahead more than a single token or backtracking when
resp-text-codefails to parse. In this case, I think backtracking is the best approach. We don't need to worry about backtracking taking exponential time, because the fallback (text) is exceptionally simple (does not call any other productions), will parse all the way up to the next CRLF, and so nothing backtrackable can be nested.This uses
ResponseParser#current_statethat was added for #599 and relies on #600 to ensure that debug error messages are not printed when backtracking.Fixes #597.